Abstract
Objective: To identify characteristics in children with epilepsy that differ between those who did versus did not have a history of preceding febrile seizures.
Background: Febrile seizures precede epilepsy in 10 to 15% of children. Little is known about the specific types of epilepsy associated with febrile seizures.
Methods: In a community-based, prospectively identified cohort of children, the association between prior febrile seizures and characteristics of the children’s epilepsy (seizure type, epilepsy syndrome, age at onset, underlying etiology, family history) were examined for 524 of the children who were aged ≥1 year at onset of epilepsy.
Results: Seventy-three (13.9%) had febrile seizures. Children with febrile seizures were more likely to have a first-degree or a second-higher-degree relative with febrile seizures and less likely to have childhood absence epilepsy and absence seizures compared with children without febrile seizures. This was especially true for simple febrile seizures. There was no specific association with localization-related forms of epilepsy. Complex, but not simple, febrile seizures were associated with younger age at onset of epilepsy. There was no evidence that focal or prolonged febrile seizures were associated with localization-related epilepsy or temporal lobe epilepsy per se. Of the three children whose initial MRIs demonstrated hippocampal atrophy, none had a history of febrile seizures.
Conclusions: At the time of diagnosis, febrile seizures are not specifically related to temporal lobe epilepsy or localization-related epilepsy in general. A genetic component for febrile seizures is suggested by its positive associations with family history, especially for simple febrile seizures. Complex febrile seizures represent an underlying age-dependent susceptibility.
Febrile seizures are the most common type of seizure in children and affect approximately 2 to 4% of all children in the United States and western Europe.1 Higher estimates are obtained in Japan and developing countries.2,3 Although it is now acknowledged to be a generally benign syndrome,4,5 children who have experienced febrile seizures are at increased risk of developing later unprovoked seizures and epilepsy compared with children without febrile seizures. Although the risk of unprovoked seizures after febrile seizures is on the order of a few percent,6-8 it is several times higher than what is seen in the general population. By the same token, a history of febrile seizures is present in 10 to 15% of people with epilepsy or unprovoked seizures,9,10 several times higher than the 2 to 4% seen in the general population.1
Among children with febrile seizures, three factors have been well established as predictors of later epilepsy: family history of epilepsy, preexisting neurologic or developmental abnormalities, and complex febrile seizures.8 The number of febrile seizures has also been associated in some instances with the risk of epilepsy.6,7,11 Beyond that, it is difficult to obtain detailed information about many aspects of the association between febrile seizures and later epilepsy by following children from the first febrile seizure because so few develop epilepsy. Although reports from surgical series have implicated febrile seizures as at least a precursor if not a cause of mesial temporal sclerosis (MTS) and intractable temporal lobe epilepsy (TLE),12-14 this association has never been demonstrated early in the course of epilepsy. In addition, there are no data, starting from the point of diagnosis, on the epidemiology and prognosis of TLE with respect to febrile seizures in the population. The available information comes primarily from patients with refractory epilepsy who are evaluated for surgery. Thus the significance of the association between febrile seizures and refractory TLE is unclear.
The purpose of the following analyses is to determine whether, in children who have epilepsy, there are clear differences between those with and without prior febrile seizures that may shed light on the association between febrile seizures and later epilepsy. Factors that were studied include the types of epilepsy and seizures, underlying etiology, family history, and neuroimaging findings.
Methods.
A detailed description of the methods used to identify and recruit the cohort and collect information has been previously reported.15 The following provides a synopsis of the main points. Children with newly diagnosed epilepsy were prospectively identified at the time of initial diagnosis from the practices of 16 child neurologists in the state of Connecticut from January 1993 through December 1997 as well as from seven pediatricians and five adult neurologists. To be eligible, a child had to have had his first unprovoked seizure before his 16th birthday. Once informed consent was obtained, the parent of the child was interviewed, all medical records pertaining to the diagnosis of the epilepsy and any previous records pertinent to previous diagnostic evaluations and underlying etiology were reviewed, and information was extracted and coded. Seizure types and epilepsy syndromes were determined by independent review of each record by three pediatric neurologists with training in clinical neurophysiology (S.R.L., F.M.T., S.S.). The International Classification of Seizures5 and of Epilepsy Syndromes16 provided the standards for making these determinations. Neurologic abnormalities were identified from the information reported in the medical record. EEG and MRI or CT interpretations, as reported in the medical record, were used for study purposes unless they were considered inadequate. Original tracings of several EEGs and a few original MRIs were obtained and re-reviewed in these instances when possible.
Definitions for etiology are those described in the International League Against Epilepsy (ILAE) Commission’s report on standards for epidemiologic research.17 The term “idiopathic” specifically refers to children who have an idiopathic form of epilepsy and no underlying neurologic abnormalities. “Cryptogenic” refers to those with other forms of epilepsy and no known underlying neurologic abnormalities. “Remote symptomatic” is used for epilepsy that occurs in the presence of a known, recognized underlying neurologic condition associated with an increased risk of epilepsy. The syndromic system recognizes similar distinctions with some exceptions. For example, a particular patient may be classified as having symptomatic localization-related epilepsy by virtue of a localized EEG focus but be of cryptogenic etiology according to the epidemiologic classification of etiology. For this reason, etiology per the epidemiologic criteria and syndromes per the international classification were classified and analyzed separately.
Febrile seizures were defined according to current guidelines.4,17 These guidelines do not specify an upper age limit for first febrile seizures other than that they occur in childhood. As in a previous study,18 we adopted an upper age limit of 10 years, thereby excluding adolescents. A febrile seizure is one that occurs in the presence of a fever (temperature ≥101 °F, 38.3 °C) in a child with no previous unprovoked seizures. Seizures with fever after the first unprovoked seizure were not considered febrile seizures.4 Intracranial infection as the underlying illness or severe metabolic abnormalities during the acute illness precluded classification of an event as a febrile seizure. A complex febrile seizure had one or more of the following three characteristics: focal onset; multiple seizures within an acute illness episode; a duration of more than 10 minutes.6,18 By necessity, information about febrile seizures was collected retrospectively. Original emergency department records and neurologists’ notes for prior febrile seizures were reviewed whenever they were available.
For this particular analysis, children were excluded if their first unprovoked seizure occurred before the age of 12 months. This was done to exclude children who developed epilepsy before they could be expected to have had a reasonable chance of having a febrile seizure. A similar approach was taken in another study of this issue.9
Analyses were performed with standard statistical methods for bivariate comparisons. Multivariable analyses were performed with logistic regression. The main analysis compared all children with prior febrile seizures to children without febrile seizures. Separate analyses compared children without febrile seizures children with a) only simple febrile seizures and b) any complex febrile seizures.
Results.
A total of 613 children were recruited into this cohort. Eighty-five (14%) experienced their first unprovoked seizure before age 12 months. A history of febrile seizures was unavailable for four children who were adopted or in foster care. This presentation is based on the remaining 524 children.
Seventy-three children (13.9%) had a history of febrile seizures. The median age at the time of the first febrile seizure was 18 months. Four children were older than 5 years at the time of their first febrile seizure, the oldest had just turned 9. Of these 73 children, 36 (49.3%) had one febrile seizure only, 15 (20.5%) had two febrile seizures, 6 (8.2%) had three febrile seizures, 16 (21.9%) had four or more febrile seizures. Thirty-nine children (53.4%) had simple febrile seizures only, and 33 (45.2%) had at least one complex febrile seizure. One child had a definite history of febrile seizures; however, details of the seizures were inadequate to determine the presence of complex features. Focal seizures occurred in 17 children, multiple seizures during a single illness episode in 19 children, and prolonged seizures (≥10 minutes) in 9 children, 7 of whom had febrile status epilepticus (≥30 minutes).17 More then one complex feature could occur in a child’s history.
Compared with children who had no prior febrile seizures (table 1), the febrile seizure group was younger at the time of the first unprovoked seizure (mean 5.5 years versus 6.9 years, p = 0.003). Boys were only slightly more likely than girls to have a history of febrile seizures. Children of Hispanic origin were also more likely to have had febrile seizures. More children with cryptogenic etiology had febrile seizures (17.9%) than those with either idiopathic (8.8%) or remote symptomatic (12.4%) etiology (p = 0.02). A family history of febrile seizures in a first-degree or a second- or higher-degree relative and a history of epilepsy in a first-degree relative were all associated with prior febrile seizures (table 1).
Bivariate associations with history of prior febrile seizures
Imaging studies were performed in 407 children (78%), and 317 had MRIs. MRIs were performed in 76% of children with symptomatic localization-related epilepsy and in 81% of the subgroup with TLE. Definite evidence of hippocampal atrophy was rare and was present in only three children. None had a history of febrile seizures. Febrile seizures were about as common in children with versus those without equivocal evidence of hippocampal atrophy or of other potentially related temporal lobe abnormalities (see table 1).
Absence seizures (typical and atypical have been combined) were relatively unlikely to have been preceded by febrile seizures compared with all other seizure types together. Beyond that, few if any positive or negative associations are apparent between seizure type and febrile seizures (table 2). Children with partial seizures, and with complex partial seizures in particular, were not more likely to have had febrile seizures compared with other children.
History of prior febrile seizures by type of seizures at onset of epilepsy
At the most basic level of the syndromic classification system, substantial differences are seen between the “not clearly focal or generalized” category (24% had febrile seizures) compared with the localization-related (14%) and generalized categories (9%, p = 0.02) (table 3). At the second level, only 8% in the primary generalized category had prior febrile seizures compared with 17% in the cryptogenic–localization-related group. Of all the primary generalized syndromes, childhood absence stands out as having the lowest proportion of children with prior febrile seizures (4%). By contrast, 25% of the “other primary generalized” group had febrile seizures. The other types of primary generalized syndromes are relatively uncommon, and therefore the estimates of their associations with prior febrile seizures are highly unstable. The proportion of children with TLE who had prior febrile seizures was similar to that in children with other types of epilepsies overall.
History of prior febrile seizures and electroclinical syndrome identified at diagnosis of epilepsy
Multivariable analysis (table 4).
Seizure syndromes are, to a significant extent, defined by seizure type(s) and age at onset. Consequently, the analysis had to consider these three factors simultaneously and determine which factors best explained the occurrence of prior febrile seizures.
Multivariable models
In the overall group (N = 524), age at first unprovoked seizure, a history of febrile seizures in a first-degree relative, a history of febrile seizures in a second- or higher-degree relative, and childhood absence epilepsy were significantly and independently associated with febrile seizures. In the presence of these four factors, a family history of epilepsy in a first-degree relative was only marginally significant (see table 4a). Seizure type, etiology, and other aspects of the epilepsy syndrome did not enter the model. Age was treated as a linear factor in all models because there was no significant departure from linearity in its association with febrile seizures. The terms for both Hispanic race and for epilepsy of undetermined onset dropped out after adjustment for other factors. An alternate model contained a term for absence seizures, instead of childhood absence epilepsy, and family history of epilepsy in a first-degree relative and was highly comparable (based on the log likelihood) with the one presented in table 4, part A.
Separate analyses were performed for children who had only simple febrile seizures versus no febrile seizures and any complex febrile seizures versus no febrile seizures.
Simple febrile seizures only (see table 4, part B).
History of febrile seizures in a first-degree relative as well as in a second- or higher-degree relative and history of epilepsy in a first-degree relative were associated with simple febrile seizures. Childhood absence epilepsy was negatively associated. The term for absence seizures did not achieve statistical significance, and the magnitude of the association was substantially less than it was for the syndrome of childhood absence epilepsy. Age at onset of epilepsy was not significantly or substantially associated with simple febrile seizures.
Complex febrile seizures (see table 4, part C).
Younger age at onset of epilepsy and a history of febrile seizures in a first-degree relative were associated with complex febrile seizures. Absence seizures were independently and negatively associated. Childhood absence epilepsy per se was not statistically significant after adjustment for these other factors.
In none of these analyses did TLE stand out as a substantial or significant correlate of either simple or complex febrile seizures. In addition, all analyses were repeated after excluding the four children who were aged 6 years and older at the time of their first febrile seizure. All estimates of odds ratios stayed in the same direction and were essentially unchanged in magnitude. The interpretation of the results was unaffected.
Children with prior febrile seizures.
Within the group of children with prior febrile seizures (n = 72 with adequate history to classify complex features), we explored possible associations between epilepsy syndromes and types of complex features. Complex features were less common in children with generalized (31%) and localization-related forms of epilepsy (42%) compared with un-determined forms of epilepsy (71%) (p = 0.07). There were minimal and ultimately nonsignificant differences seen for focal and prolonged febrile seizures. However, multiple febrile seizures (within an illness) occurred in 10 of 45 of those with localization-related epilepsy (23%), none of 9 with generalized epilepsy (0%), and 9 of 14 (64%) with undetermined-onset epilepsy (p < 0.001).
In children who had localization-related epilepsy (n = 45), focal febrile seizures were equally likely to have occurred in those with idiopathic and symptomatic localization-related epilepsy. Prolonged febrile seizures occurred in none of 6 children with idiopathic epilepsy (0%), 1 of 15 with cryptogenic epilepsy (7%), and 5 of 24 with symptomatic localization-related epilepsy (21%) (p = 0.18). Multiple febrile seizures occurred in none of the children with idiopathic localization-related epilepsy and in 25% of each of the other two groups. The pattern of findings was similar for TLE alone (a subgroup of the symptomatic localization-related category) with the exception that only multiple (not prolonged or focal) febrile seizures in one illness were more common in TLE.
In children with prior febrile seizures, recurrent febrile seizures (during different illness episodes) were most common in the symptomatic localization-related group (62.5%). Seventy percent of the TLE group and 80% of the “symptomatic localization-related by virtue of etiology” group had recurrent febrile seizures. In addition, 64% of the undetermined group had recurrent febrile seizures. By contrast, only approximately one-third of each of the other syndrome groups had recurrent febrile seizures.
Discussion.
The frequency of febrile seizures in this cohort (13.9%) and the associations with family history have been observed in other contexts.9,19-21 It is therefore not surprising that family history of febrile seizures would be associated with febrile seizures within a cohort of children with epilepsy. Its association with both simple and complex febrile seizures in a setting where all patients have epilepsy provides further evidence for the specificity of febrile seizures as a genetic syndrome whose relationship to epilepsy may represent an age-dependent and somewhat broad-based susceptibility to seizures.22 Genetic models provide evidence of separate polygenic and single-gene mechanisms.23
The strong negative correlation with childhood absence or perhaps any form of epilepsy associated with absence seizures may also be related to the underlying genetics. The statistical models that were derived did not provide an unequivocal answer as to whether the specific syndrome or the seizure type was more important. As would be expected, these two factors overlapped considerably. In a series of children from Japan with febrile seizures who developed epilepsy, there was a noticeable paucity of children with absence seizures.24 This may mean that the mechanisms involved in childhood absence epilepsy or possibly in any form of epilepsy associated with absence seizures are very different from those involved in febrile seizures.
The Canadian study also reported that children with epilepsy preceded by febrile seizures had a younger age at onset than those with no prior febrile seizures.9 In our study, the association with age was entirely limited to complex febrile seizures and suggests that complex febrile seizures are an age-dependent marker for susceptibility to epilepsy.
Febrile seizures and TLE, MTS, and complex partial seizures.
The relationship between febrile seizures in childhood, complex partial seizures, TLE, and MTS in particular has long been assumed based upon strong associations seen among these and other factors in adults12,25 as well as in children,26-28 largely with refractory epilepsy.
A recent study examined these issues in a mixed epilepsy population of all ages, not all of whom had long-standing epilepsy.10 This study also found a strong association between prior febrile seizures and temporal but not extra-temporal lobe epilepsy.
One recent report from a small case series of children with febrile status epilepticus indicated that hippocampal atrophy was sometimes already present before a prolonged febrile seizure.29 In a few cases, early evidence consistent with MTS could be observed immediately after prolonged febrile seizures. By contrast, another preliminary report of 13 patients with complex febrile seizures followed for 2 years with repeat MRIs indicated that any evidence of hippocampal atrophy was already present at the time of the initial febrile seizure.30
One child in our study had prolonged seizures in the context of viral encephalitis. This immediately evolved into epilepsy. Early MRI changes in the mesial temporal lobe during the acute encephalitis were apparent and evolved into hippocampal atrophy on subsequent imaging. The presence of encephalitis, however, precludes the event from being considered a “febrile seizure.”4,17 Because histories of prior provoked seizures in surgical candidates are typically taken after many years, this raises the possibility of misclassification of some events as febrile seizures when, in fact, a more serious process may have been involved.
In light of the findings from these studies, it is notable that, in our cohort of children with newly diagnosed epilepsy, there was no evidence that prior febrile seizures were associated with complex partial seizures, localization-related epilepsy, or specifically with TLE or with hippocampal atrophy or other potentially related findings in the temporal lobe. This was true even when analyses were limited to children whose first febrile seizures occurred when the children were younger than age 5 or even younger than age 3, the range in which there is the greatest concern about the possible effects of febrile seizures.31 Also, none of the three children with MRI-documented hippocampal atrophy had a history of febrile seizures. Although the evidence of MTS may require time to develop before it is detectable on MRI, these children were usually imaged at the onset of epilepsy and after some time had elapsed since their febrile seizure(s).
Cohort studies of newly diagnosed epilepsy.
At least two other large studies have examined the differences in childhood-onset epilepsy preceded versus not preceded by febrile seizures.9,32 A third study explicitly examined the relationship between febrile seizures and TLE.33 These studies were done before the publication of the current ILAE classification of epileptic syndromes16 and the epidemiologic criteria for epilepsy research.17 In addition, for these earlier studies, neuroimaging was not as widely available, if at all. By contrast, the majority of children in our study with nonidiopathic localization-related epilepsy had an MRI. This limits some direct comparisons between studies. Regardless, all available studies (total N\S2,000) clearly agree on the notable lack of an association between febrile seizures and partial seizures, localization-related epilepsy, and TLE per se at the time of diagnosis of epilepsy in children. One other recent study used MRS to examine metabolic markers of hippocampal sclerosis in a small series of children with epilepsy.34 Although abnormalities were found, the authors reported no differences between those with versus without a history of complex febrile seizures.
Other clinical data.
Other observations present challenges to a causal explanation for the association between febrile seizures and MTS and TLE. A recent report of a series of surgical patients demonstrated that the association between febrile seizures and MTS was largely explained by the age at onset of epilepsy.35 Specifically, both MTS and febrile seizures were correlated with younger age at onset of epilepsy. After adjustment for age at onset, febrile seizures, and particularly prolonged febrile seizures, were no longer associated with MTS. The authors suggested that the relationship between febrile seizures and MTS most likely represents preexisting susceptibility or underlying pathology and that one does not cause the other.
Data from three randomized clinical trials indicate that early treatment can prevent recurrent febrile seizures but that this does not translate into a decreased risk for later unprovoked seizures and epilepsy.36-40 If removal of the putative cause has no impact on the putative effect, it calls into question whether one factor truly causes the other. This suggests that a substantial part of the risk is inherent in the individual and not in the febrile seizures per se. The findings of other pathologies, such as migrational defects, in the temporal lobes of individuals with MTS further suggest this possibility.41,42
Animal models.
Evidence from animal models of febrile seizures in the developing brain also fails to demonstrate a long-lasting impact of early seizures on the hippocampal formation and the limbic system in general.43,44 Further evidence tends to support the role of underlying pathology by demonstrating that the presence of experimentally induced neuro-migrational defects (NMDs) decreases the threshold for hyperthermic seizures45 and that, relative to controls, rat pups with NMDs have a lower threshold for kindling and are more susceptible to hippocampal neuronal damage.46
Exploration limited to those children in the cohort who had febrile seizures indicated that multiple and prolonged febrile seizures as well as recurrent febrile seizures were more common in children with symptomatic and cryptogenic compared with idiopathic localization-related epilepsy. For TLE this was seen only for multiple (in one illness) and recurrent febrile seizures and not for focal and prolonged febrile seizures. The same was true of epilepsy of undetermined onset, much of which may, in fact, be localization-related epilepsy. Localization in our series was based largely upon routine scalp EEG and was not as precise as that available in surgical series. Nonetheless, our findings were simply not supportive of the clear associations seen in surgical series between TLE and a history of focal or prolonged febrile seizures. Multivariable models also revealed no association between febrile seizures and localization-related epilepsy.
This study has further explored the relationship between febrile seizures and epilepsy by examining the types of epilepsy syndromes and seizures associated with febrile seizures in children at the time of their initial diagnosis of epilepsy. Although febrile seizures are not a predictor of outcome in childhood-onset epilepsy,32,47,48 it would seem incontrovertible that there is an association between febrile seizures and MTS in patients with refractory TLE. The fact that such an association is not present at the time epilepsy is diagnosed strongly argues against a causal association between febrile seizures and TLE at least for the vast majority of cases. Even large epidemiologic studies do not have the power to address whether a causal association might exist in extremely rare cases of febrile seizures that last in excess of 2 hours.31 Other data, reviewed above, further argue against a causal association. To explain the disparities in epidemiologic studies versus surgical series, it is necessary to posit a more complicated relationship between febrile seizures and intractable TLE. For example, there may be an interaction between the seizures and underlying abnormalities in certain instances, which results in MTS. Alternatively, febrile seizures do not selectively affect the temporal lobe, but when they do, the overall long-term outcome of those who develop epilepsy may be poor. Finally, febrile seizures, particularly complex febrile seizures, may simply be an age-dependent marker or expression of preexisting pathologies whose significance is not appreciated until later.
Long-term follow-up of this cohort will provide further information about the relative prognostic significance of prior febrile seizures, etiology, seizure types, and syndromes with respect to the seizure outcomes of childhood-onset epilepsy.
Acknowledgments
Supported by a grant from the National Institutes of Health—NINDS RO1-NS31146.
Acknowledgment
The authors thank the physicians in Connecticut who referred patients for study: Drs. Robert Cerciello, Francis Dimario, Barry Russman, Michelle Kleiman, Carol Leicher, Edwin Zalneraitis, Philip Brunquell, Laura Ment, Edward Novotny, Bennet Shaywitz, S. Nallainathan, Alok Bhargava, Martin Kreminitzer, Barbara Coughlin, Harriet Fellows, Jack Finkelstein, Daniel Moalli, Louise Resor, Brenda Balch, Patricia Braun, Owen Erlich, Bernard Giserman, John Monroe, Lawrence Rifkin, Lourdes Rosales, and Murray Engel. The authors thank Drs. Edward Novotny and Francis DiMario for reinterpreting selected EEGs. Dr. Eugene Shapiro facilitated many administrative issues for us. The authors thank the research associates, Susan Smith-Rapaport, Barbara Beckerman, Heather LaCoste, Lynnette Bates, Joann Gehrels, and Kris Engel for their dedicated work on this project and Wuthikrai Uayingsak for his exceptional programming expertise. This study was made possible by the generous participation of the many parents and their children in this study.
- Received January 14, 1999.
- Accepted June 22, 1999.
References
Letters: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hemiplegic Migraine Associated With PRRT2 Variations A Clinical and Genetic Study
Dr. Robert Shapiro and Dr. Amynah Pradhan